exp010: apxJ8
Browse files
experiments/exp_010_component_potentials/runs/apxJ8/checkpoint.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8457ea916a6a7f1474b114ca94e37b7060fae2cd6de23767bd6aebace685b130
|
| 3 |
+
size 28998010
|
experiments/exp_010_component_potentials/runs/apxJ8/result.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run": "apxJ8",
|
| 3 |
+
"family": "headgeo",
|
| 4 |
+
"dev_final": 0.01145327091217041,
|
| 5 |
+
"rank_final": 3.9459526538848877,
|
| 6 |
+
"statute_final": "uniform",
|
| 7 |
+
"params": 7149387,
|
| 8 |
+
"minutes": 7.3,
|
| 9 |
+
"bpb": 2.8962410404963643,
|
| 10 |
+
"bpb_eval": NaN,
|
| 11 |
+
"coverage": 0.8077392578125,
|
| 12 |
+
"gate_acc": 0.8551025390625,
|
| 13 |
+
"mode_spread_deg": 86.97556741923984,
|
| 14 |
+
"mix_entropy": 0.9641604234794341,
|
| 15 |
+
"write_kl": 0.13950371742248535,
|
| 16 |
+
"kernel_aux": NaN,
|
| 17 |
+
"loss": 6.0365142822265625,
|
| 18 |
+
"condition": {
|
| 19 |
+
"tag": "apxJ8",
|
| 20 |
+
"dim": "384",
|
| 21 |
+
"n_layers": "4",
|
| 22 |
+
"n_heads": "6",
|
| 23 |
+
"K": "64",
|
| 24 |
+
"seq_len": "256",
|
| 25 |
+
"batch_size": "32",
|
| 26 |
+
"accum_steps": "1",
|
| 27 |
+
"d_point": "48",
|
| 28 |
+
"head": "hybrid",
|
| 29 |
+
"steps": "5000",
|
| 30 |
+
"log_every": "500",
|
| 31 |
+
"bank_scorer": "apmix",
|
| 32 |
+
"n_pointers": "8",
|
| 33 |
+
"bank_source": "atlas/bank_16384_tri.pt"
|
| 34 |
+
}
|
| 35 |
+
}
|
experiments/exp_010_component_potentials/runs/apxJ8/snapshots.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b19971716cb0bea5e05e5bc5532caf3ffde0c370f44d395f436879b9ac755f3
|
| 3 |
+
size 17188
|